JP2020045024A - Electronic control unit - Google Patents

Abstract

An object of the present invention is to evaluate the reliability of a trajectory planning calculation with a low load without multiplexing the calculation or using an additional sensor.
An electronic control unit acquires information about a periphery of a vehicle as sensor information from a plurality of sensors in each processing cycle, and integrates the acquired sensor information to create vehicle surrounding information in each processing cycle. A trajectory planning unit 114 that calculates a planned trajectory that the vehicle will travel in the future using the vehicle surrounding information for each processing cycle; and a trajectory evaluation unit 123 that evaluates the reliability of the trajectory planning unit. The position of the vehicle in the second processing cycle, which is a processing cycle after the first processing cycle, in the planned trajectory calculated by the trajectory planning unit 114 in the first processing cycle, and the integration unit in the second processing cycle. The reliability of the trajectory planning unit 114 in the first processing cycle is evaluated using the vehicle surrounding information created by the 112.
[Selection diagram] FIG.

Description

  The present invention relates to an electronic control device.

In order to realize an advanced automatic driving system, the automatic driving ECU, which is a higher-level control device that controls automatic driving, has a certain period of time until the driver takes over the operation even if a failure occurs in the automatic driving system. It is required to continue the operation. Examples of such a failure include, for example, an abnormality that occurs during an operation on an arithmetic processing device that performs an operation for automatic driving control, and an abnormality that occurs in a sensor. In order to realize the continuation of the operation for a certain period of time, it is necessary to detect these abnormalities and switch to control corresponding to the abnormalities. In order to detect such failures and abnormalities, there are generally a method of multiplexing arithmetic processing and sensors and comparing the outputs, and a method of verifying the validity of the calculation results and sensor output values using different sensor values and calculation results. Used. Of these, multiplexing has problems such as a large-scale system complexity and an increase in calculation load due to an increase in the number of sensors, and a method for performing validity verification is required.
Patent Literature 1 includes an input member that moves forward and backward by operating a brake pedal, an assist member that is arranged to be relatively movable to the input member, and an electric actuator that moves the assist member forward and backward. In an electric booster for generating a brake hydraulic pressure boosted in a master cylinder by an assist thrust applied to the assist member in response to the movement of the input member, the electric booster detects an absolute displacement amount of the input member. Any one of input absolute displacement detection means, relative displacement detection means for detecting the relative displacement between the input member and the assist member, or assist absolute displacement detection means for detecting the absolute displacement of the assist member And a phase between the input member and the assist member in accordance with a detection signal of the input absolute displacement amount detecting means. A target displacement amount at which the displacement relationship is variable is set, and based on a signal from the relative displacement amount detecting means or the assist absolute displacement amount detecting means, the relative displacement relationship between the input member and the assist member is set to the target displacement amount. There is disclosed an electric booster characterized by comprising control means for controlling the electric actuator so that

JP 2010-215234 A

  In the invention described in Patent Document 1, the cost of evaluating the reliability of the trajectory planning calculation is high.

  An electronic control device according to a first aspect of the present invention acquires information on a periphery of a vehicle as sensor information from a plurality of sensors in each processing cycle, integrates the acquired sensor information, and integrates the vehicle peripheral information in each processing cycle. An integration unit that creates a trajectory planning unit that calculates a planned trajectory that the vehicle will travel in the future using the vehicle surrounding information for each processing cycle, and a trajectory evaluation unit that evaluates the reliability of the trajectory planning unit. The trajectory evaluation unit includes a position of the vehicle in a second processing cycle that is a processing cycle after the first processing cycle in the planned trajectory calculated by the trajectory planning unit in a first processing cycle. And the vehicle surrounding information created by the integration unit in the second processing cycle, and evaluates the reliability of the trajectory planning unit in the first processing cycle.

  According to the present invention, the reliability of the trajectory planning calculation can be evaluated with a low load without multiplexing the calculation and using an additional sensor.

Overall configuration diagram of the automatic driving system S The flowchart which shows the outline of the automatic operation function realized by the microcomputer 21 The figure which shows the outline | summary of the sensor fusion shown in step S101 of FIG. Conceptual diagram of vehicle surrounding information 907 Conceptual diagram of risk map 908 Conceptual diagram of the planned trajectory 909 Functional block diagram of microcomputer 21 The figure which shows the information which the 1st evaluation part 121, the 2nd evaluation part 122, and the 3rd evaluation part 123 use. The figure which shows the determination content of the 1st determination part 131. The figure which shows the determination content of the 2nd determination part 132

-Embodiment-
Hereinafter, an embodiment of an electronic control unit according to the present invention will be described with reference to FIGS.

  FIG. 1 is an overall configuration diagram of an automatic driving system S including an electronic control device according to the present invention. The automatic driving system S includes a sensor group 1, an automatic driving ECU 2, a command unit 25, a lower ECU group 3, and a vehicle position sensor 14. The automatic driving system S is mounted on a vehicle, and a vehicle on which the automatic driving system S is mounted is hereinafter referred to as “own vehicle”.

  The sensor group 1 includes two sensors, and the two sensors are referred to as a first sensor 11 and a second sensor 12 in the present embodiment. These two sensors are, for example, a millimeter wave radar and a camera. The first sensor 11 and the second sensor 12 acquire information on the surroundings of the host vehicle and output the information to the automatic driving ECU 2. Hereinafter, information about the surroundings of the host vehicle that is output from the first sensor 11 and the second sensor 12 to the automatic driving ECU 2 is also referred to as “sensor data” or “sensor information”. The sensor data includes, as a time stamp, the time at which each sensor acquired the data, in other words, the time at which sensing was performed.

  The own vehicle position sensor 14 is a sensor that calculates the position of the own vehicle, for example, latitude and longitude. The vehicle position sensor 14 is, for example, a GPS receiver, and receives radio waves from a plurality of satellites constituting the satellite navigation system, and calculates the position of the vehicle by analyzing signals included in the radio waves. The vehicle position sensor 14 outputs the calculated latitude and longitude to the automatic driving ECU 2.

  The automatic driving ECU 2 is an electronic control unit (Electronic Control Unit). The automatic driving ECU 2 acquires the map information 13 and uses it for the calculation described later. The map information 13 may be obtained from a device provided in the host vehicle, or may be obtained by communication from outside the host vehicle. The automatic driving ECU 2 reads the map information 13 from, for example, a nonvolatile memory mounted on the host vehicle, for example, a flash memory.

  The automatic driving ECU 2 includes a microcomputer 21, and the microcomputer 21 performs a calculation relating to automatic driving of the own vehicle. The microcomputer 21 includes a CPU serving as a central processing unit, a ROM serving as a read-only storage device, and a RAM serving as a readable and writable storage device. The CPU expands a program stored in the ROM into the RAM and executes the program. The functions described below are realized. However, the microcomputer 21 may be realized by an FPGA (Field Programmable Gate Array), which is a rewritable logic circuit, or an ASIC (Application Specific Integrated Circuit), which is an application-specific integrated circuit, instead of a combination of a CPU, a ROM, and a RAM. Good. The microcomputer 21 may be realized by a combination of different configurations, for example, a combination of a CPU, a ROM, a RAM, and an FPGA, instead of the combination of the CPU, the ROM, and the RAM.

  The computation performed by the microcomputer 21 uses the information on the surroundings of the own vehicle output by the sensor group 1, the map information 13, and the position of the own vehicle output by the own vehicle position sensor 14. The microcomputer 21 calculates a future trajectory in which the host vehicle will travel, and outputs the calculated trajectory to the command unit 25. The command unit 25 is, for example, an electronic control unit (Electronic Control Unit). However, the command section 25 may be included in the automatic driving ECU 2. The command unit 25 outputs a specific operation command to each device included in the lower ECU group 3 based on the future trajectory of the vehicle output by the microcomputer 21.

  The lower ECU group 3 controls the own vehicle based on the output of the automatic driving ECU 2. The lower ECU group 3 includes, for example, a brake control device 31 for controlling a brake, an engine control device 32 for controlling an engine, and a steering control device 33 for controlling steering. In addition, the sensor information input from the sensor group 1 to the automatic driving ECU 2 is not limited to the case where the raw data directly output from each sensor is used, or the data which is pre-processed by the ECU mounted on each sensor. Is also good.

  FIG. 2 is a flowchart showing an outline of the automatic driving function realized by the microcomputer 21. First, in step S101, the microcomputer 21 performs a process of fusing sensor fusion, that is, the output of the first sensor 11 and the output of the second sensor 12. By this sensor fusion processing, vehicle surrounding information 907 which is a map around the own vehicle is obtained. In subsequent step S102, the microcomputer 21 performs an action prediction for predicting the action of the peripheral object using the own vehicle surrounding map obtained in step S101, and creates a risk map 908 to be described later. Then, in the subsequent step S103, the microcomputer 21 performs an own-vehicle trajectory plan for generating a trajectory of the own vehicle using the calculation result of the step S102, and generates a planned trajectory 909.

  Thereafter, the microcomputer 21 determines whether or not the operation of the automatic driving ECU 2 is to be terminated. If it is determined that the operation is to be terminated, the microcomputer 21 terminates the processing illustrated in FIG. 2 and determines that the operation of the automatic driving ECU 2 is not to be terminated. Returns to step S101. For example, the microcomputer 21 makes an affirmative decision in step S104 when the ignition switch of the own vehicle is turned off, and otherwise makes a negative decision in step S104. If a negative determination is made in step S104, the microcomputer 21 executes steps S103 to S104 at predetermined processing cycles, for example, at every 100 ms. For example, when the processing cycle is 100 ms, the sensor fusion process in step S101 is performed, and the next step S101 is performed again after 100 ms.

  FIG. 3 is a diagram showing an outline of the sensor fusion shown in step S101 of FIG. The time synchronization processing unit 111 and the integration unit 112 shown in FIG. 2 are functional blocks showing functions realized by the microcomputer 21. In the sensor fusion, first, a time synchronization process is executed by the time synchronization processing unit 111. Since the sensor data output by the first sensor 11 and the second sensor 12 are not time-synchronized, the time synchronization processing 41 generates sensor data at a time synchronized with the processing cycle.

  Specifically, the microcomputer 21 performs extrapolation and interpolation in a time series using the time-series sensor data received from the first sensor 11, and performs synchronization, which is sensor data of the first sensor 11 at a predetermined time. And outputs the completed first sensor data 901. Similarly, the microcomputer 21 performs extrapolation and interpolation in time series using the time-series sensor data received from the second sensor 12, and outputs the synchronized second data, which is the sensor data of the second sensor 12 at a predetermined time. Two sensor data 902 is output.

  The predetermined time is a time at which the processing cycle has elapsed from the time of the previous synchronization. That is, the synchronized first sensor data 901 and the synchronized second sensor data 902 are the sensor data of the first sensor 11 and the second sensor 12 at the same time obtained by calculation. Note that, here, although simply referred to as “synchronized first sensor data 901” or “synchronized second sensor data 902”, since the situation around the own vehicle changes, the value is different for each processing cycle. . Details will be described later.

  The integrated first sensor data 901, the synchronized second sensor data 902, the map information 13, and the own vehicle position are input to the integration unit 112. The integrating unit 112 integrates the synchronized first sensor data 901 and the synchronized second sensor data 902 with the map information 13 to generate vehicle surrounding information 907. The vehicle surrounding information 907 includes not only the position of an object existing around the own vehicle but also the surrounding static information obtained from the map information 13 based on the own vehicle position input by the own vehicle position sensor 14, such as an intersection or the like. The position of the pedestrian crossing, the number of driving lanes, etc. are included. The position of an object existing around the host vehicle may be calculated and output based on information obtained by sensing each of the first sensor 11 and the second sensor 12, or the output of each sensor may be used. For example, the time synchronization processing unit 111 of the automatic driving ECU 2 may calculate based on the calculation.

(Conceptual diagram of vehicle surrounding information 907)
FIG. 4 is a conceptual diagram of the vehicle surrounding information 907. FIG. 4 shows, in addition to the stationary objects around the own vehicle obtained from the map information 13, the positions and sizes of the objects around the own vehicle detected by the first sensor 11 and the second sensor 12. Specifically, FIG. 4 shows a host vehicle 71, an oncoming vehicle position 72, a stopped vehicle 73, and a pedestrian position 74. Although FIG. 4 illustrates the own vehicle 71 for explanation, the vehicle surrounding information 907 does not include the position information of the own vehicle 71.

(Conceptual diagram of risk map 908)
FIG. 5 is a conceptual diagram of the risk map 908. The risk map 908 is information indicating the position of an object existing around the own vehicle. As shown in FIG. 5, the risk map 908 includes the predicted positions of the oncoming vehicle predicted position 82, the stopped vehicle predicted position 83, and the walking predicted position 84. However, in FIG. 5, information of the vehicle surrounding information 907 shown in FIG. 4 is indicated by broken lines to facilitate understanding. Although FIG. 5 shows only one predicted position of each object, a plurality of predicted positions may be included. The number of predicted positions is, for example, a number necessary for the trajectory planning process. For example, when the own vehicle trajectory is planned for a period corresponding to 10 seconds ahead every 100 milliseconds in the trajectory planning process, a maximum of 100 predicted positions in each object are generated.

(Conceptual diagram of the planned trajectory 909)
FIG. 6 is a conceptual diagram of the planned trajectory 909. The planned trajectory 909 is information indicated by a thick solid line in FIG. 6 and is a trajectory where the own vehicle is planned to run in the future. Using the planned trajectory 909, the command unit 25 calculates the position of the host vehicle for each processing cycle by further using information on the speed of the host vehicle and the known processing cycle. For example, the position of the host vehicle in the most recent future indicated by reference numeral 91, that is, the immediately following processing cycle, and the position of the host vehicle in the future indicated by reference numeral 92 can be calculated.

(Function block of microcomputer 21)
FIG. 7 is a functional block diagram showing the functions of the microcomputer 21. However, since the sensor group 1 and the command unit 25 in the configuration shown in FIG. 7 exist outside the microcomputer 21, they are indicated by broken lines. The microcomputer 21 includes, as its functions, a time synchronization processing section 111, an integration section 112, an action prediction section 113, a trajectory planning section 114, an action prediction storage section 115, a trajectory plan storage section 116, a first evaluation section 121 , A second evaluation unit 122, a third evaluation unit 123, a first determination unit 131, a second determination unit 132, and a mode transition unit 141. Hereinafter, the first evaluation unit 121 may be referred to as a prediction evaluation unit, the second evaluation unit 122 may be referred to as an integrated evaluation unit, and the third evaluation unit 123 may be referred to as a trajectory evaluation unit.

  Hereinafter, for convenience, the N-th processing cycle is referred to as “processing cycle N”, the next processing cycle is referred to as “processing cycle N + 1”, and the processing cycle immediately before the processing cycle N is referred to as “processing cycle N−1”. .

  The time synchronization processing unit 111 calculates the synchronized first sensor data 901 and the synchronized second sensor data 902, as described with reference to FIG. The time synchronization processing unit 111 outputs both the calculated synchronized first sensor data 901 and the synchronized second sensor data 902 to the integration unit 112 and the first evaluation unit 121. The integration unit 112 calculates the vehicle surrounding information 907 as described with reference to FIG. The integration unit 112 outputs the calculated vehicle surrounding information 907 to the behavior prediction unit 113, the second evaluation unit 122, and the third evaluation unit 123.

  The behavior predicting unit 113 predicts the future behavior of an object existing in the vicinity using the vehicle peripheral information 907 and outputs the predicted behavior as a risk map 908. The behavior predicting unit 113 predicts behavior at a plurality of future times, and for example, in the processing cycle N, predicts the position of the object in the processing cycle N + 1, the processing cycle N + 2, the processing cycle N + 3,. The behavior prediction unit 113 may accumulate the vehicle surrounding information 907 output from the integration unit 112, for example, and predict a future position of the object from a change in the position in a time series. Further, the behavior predicting unit 113 may predict the future position of the object by using the output of the sensor group 1 and using the speed information of the object (not shown). Further, the behavior prediction unit 113 may perform detailed prediction by a method such as correcting the route by regarding the interaction between the objects as the repulsive force between the objects. The behavior prediction unit 113 outputs the risk map 908 to the trajectory planning unit 114 and the behavior prediction storage unit 115.

  The trajectory planning unit 114 uses the risk map 908 to calculate the route on which the vehicle should travel, in other words, plans the trajectory. The trajectory planning unit 114 may plan the trajectory by referring to information of nodes such as a destination of the vehicle and an intersection to be reached next. The trajectory planning unit 114 may calculate a trajectory for preventing a collision based on, for example, a distance to a surrounding object, or may calculate a trajectory such that a travel time of the own vehicle to a destination is shortened. The trajectory may be calculated such that the acceleration affecting the ride comfort of the occupant is reduced. The trajectory planned by the trajectory planning unit 114 also includes the position information of the own vehicle in the next processing cycle. The trajectory planning unit 114 outputs the calculated planned trajectory 909 to the command unit 25 and the trajectory plan storage unit 116.

  The behavior prediction storage unit 115 temporarily stores the risk map 908 input from the behavior prediction unit 113, and stores the risk map 908 in the immediately preceding processing cycle, and the risk map 908A created in the immediately preceding processing cycle. Is output to the first evaluation unit 121 and the second evaluation unit 122. For example, in the processing cycle N, the behavior prediction storage unit 115 receives the risk map 908 created in the processing cycle N from the behavior prediction unit 113, and outputs the immediately preceding risk map 908A created in the processing cycle N-1. However, although the risk map 908 includes information on the position of the object in a plurality of future processing cycles, the immediately preceding risk map 908A may include information on the position of the object in the processing cycle N. That is, the last risk map 908A may include the position of the object in the processing cycle N predicted in the processing cycle N-1. The relationship between the processing cycle and the data will be described later in detail with reference to FIG.

  The trajectory plan storage unit 116 temporarily stores the planned trajectory 909 input from the trajectory planning unit 114, and is the planned trajectory 909 stored in the immediately preceding processing cycle and the immediately preceding planned trajectory 909A created in the immediately preceding processing cycle. Is output to the third evaluation unit 123. For example, in the processing cycle N, the trajectory plan storage unit 116 receives the risk map 908 created in the processing cycle N from the trajectory planning unit 114, and outputs the immediately preceding planned 909A created in the processing cycle N-1. However, the planned trajectory 909 contains information on the trajectory that is the position of the own vehicle in a plurality of future processing cycles, whereas the immediately preceding planned 909A only needs to include information on the position of the own vehicle in the processing cycle N. . That is, the immediately preceding planned track 909A may include the position information of the own vehicle in the processing cycle N predicted in the processing cycle N-1. The trajectory plan storage unit 116 may calculate the position of the own vehicle in the processing cycle N using the planned trajectory 909, or the third evaluation unit 123 may use the information on the trajectory to calculate the position of the own vehicle in the processing cycle N. It may be calculated.

  The first evaluation unit 121 performs an evaluation described below using the synchronized first sensor data 901, the synchronized second sensor data 902, and the immediately preceding risk map 908 </ b> A, and outputs an evaluation result to the first determination unit 131. The second evaluation unit 122 performs an evaluation described later using the vehicle surrounding information 907 and the immediately preceding risk map 808A, and outputs an evaluation result to the first determination unit 131. The third evaluation unit 123 performs an evaluation described later using the vehicle surrounding information 907 and the immediately preceding planned trajectory 909A, and outputs an evaluation result to the second determination unit 132. The first determination unit 131 performs a determination described later using the evaluation result of the first evaluation unit 121 and the evaluation result of the second evaluation unit 122, and outputs a first abnormality determination result 921. The second determination unit 132 performs a determination described below using the first abnormality determination result 921 output by the first determination unit 131 and the evaluation result of the third evaluation unit 123, and outputs a second abnormality determination result 922. Before describing the detailed operation of the first evaluation unit 121 and the like, the time-series relationship of data will be described again with reference to FIG.

  FIG. 8 is a diagram illustrating information used by the first evaluation unit 121, the second evaluation unit 122, and the third evaluation unit 123. In the example illustrated in FIG. 8, the processing in two processing cycles of the processing cycle N and the processing cycle N + 1 is described. The upper half of FIG. 8 illustrates outputs of the time synchronization processing unit 111, the integration unit 112, the behavior prediction unit 113, and the trajectory planning unit 114. The lower half of FIG. 8 illustrates information used by the first evaluation unit 121, the second evaluation unit 122, and the third evaluation unit 123. However, the information used by the first evaluator 121, the second evaluator 122, and the third evaluator 123 does not particularly indicate the origin of information generated in the same processing cycle, and indicates information generated in different processing cycles. Arrows indicate the origin only when used. However, in FIG. 8, the description of the behavior prediction storage unit 115 and the trajectory plan storage unit 116 is omitted.

  8, the synchronized first sensor data 901 calculated and output by the time synchronization processing unit 111 in the processing cycle N is described as “S1_N”, and the synchronized second sensor data 902 output in the processing cycle N is “S1_N”. S2_N. " If the calculated processing cycle is different, “N” is appropriately rewritten accordingly, and for example, the synchronized first sensor data 901 calculated and output in the processing cycle N + 1 is described as “S1_N + 1”. The fact that the suffix is changed in accordance with the calculated processing cycle is common in the present embodiment, and description thereof will be omitted below.

  In FIG. 8, the vehicle surrounding information 907 calculated by the integration unit 112 in the processing cycle N is described as “Aro_N”. In FIG. 8, the risk map 908 in the processing cycle N + 1 calculated by the action prediction unit 113 in the processing cycle N is described as “PreAro_N_N + 1”. This description is a combination of “Pre” indicating a prediction, “N” indicating a calculated processing cycle, and “N + 1” indicating a predicted processing cycle. As described above, since the behavior prediction unit 113 also calculates the risk map 908 of the processing cycle N + 2 in the processing cycle N, “PreAro_N_N + 2” is also illustrated in FIG. In FIG. 8, the position of the own vehicle in the processing cycle N + 1 included in the planned trajectory 909 calculated by the trajectory planning unit 114 in the processing cycle N is described as “PrePos_N_N + 1”.

  In the processing cycle N, the first evaluation unit 121 includes S1_N and S2_N output from the time synchronization processing unit 111 in the processing cycle N and PreAro_N-1_N output from the behavior prediction unit 113 in the processing cycle N-1. Is entered. In the processing cycle N + 1, the first evaluation unit 121 receives S1_N + 1 and S2_N + 1 output from the time synchronization processing unit 111 in the processing cycle N + 1 and PreAro_N_N + 1 output from the behavior prediction unit 113 in the processing cycle N.

  In the processing cycle N, the second evaluation unit 122 receives Aro_N, which is the output of the integration unit 112 in the processing cycle N, and PreAro_N-1_N, which is the output of the behavior prediction unit 113 in the processing cycle N-1. In the processing cycle N + 1, Aro_N + 1 that is the output of the integration unit 112 in the processing cycle N + 1 and PreAro_N_N + 1 that is the output of the behavior prediction unit 113 in the processing cycle N1 are input to the second evaluation unit 122.

  In the processing cycle N, the third evaluation unit 123 receives Aro_N, which is the output of the integration unit 112 in the processing cycle N, and PrePos_N-1_N, which is the output of the trajectory planning unit 114 in the processing cycle N-1. In the processing cycle N + 1, the third evaluation unit 123 receives Aro_N + 1 output from the integration unit 112 in the processing cycle N + 1 and PrePos_N_N + 1 output from the trajectory planning unit 114 in the processing cycle N.

  The operation of the first evaluation unit 121, the second evaluation unit 122, and the third evaluation unit 123 in the processing cycle N will be described.

  In the processing cycle N, the first evaluation unit 121 determines S1_N, which is the synchronized first sensor data 901 in the processing cycle N, S2_N, which is the synchronized second sensor data 902 in the processing cycle N, and the behavior in the processing cycle N-1. The output of the prediction unit 113 is compared with PreAro_N-1_N to detect the occurrence of an abnormality and determine the location of the abnormality. However, it is inappropriate to perform a strict match determination in the comparison operation in the first evaluation unit 121, and it is determined that they match when the difference is within a predetermined threshold. Therefore, it can be said that the first evaluation unit 121 evaluates the consistency between S1_N, S2_N, and PreAro_N-1_N.

  For example, the first evaluation unit 121 calculates the center point of the position of the obstacle included in S1_N, the position of the obstacle included in S2_N, and the center point of the position of the obstacle included in PreAro_N-1_N, and evaluates the distance from the center point. I do. In addition, the first evaluation unit 121 may perform weighting according to sensor characteristics such as a relative distance and an angle with the host vehicle in calculating the center point to improve accuracy. In this way, a match with a range is determined by comparing the obtained distance with the threshold, and a determination is made by majority based on the sensor error.

  For example, if the position of the obstacle included in S1_N substantially matches the position of the obstacle included in PreAro_N-1_N, but the position of the obstacle included in S2_N is significantly different from the other two, the first evaluation unit 121 Determines that there is a problem with the output of the second sensor 12 based on the following idea. In other words, the true position of the obstacle is not always clear, but it is determined by a 2: 1 majority rule that the outputs of S1_N and PreAro_N-1_N are more likely, and there is a problem in S2_N, that is, the output of the second sensor 12. Judge.

  The first evaluation unit 121 outputs to the first determination unit 131 the presence or absence of a match and, if there is any match, information on the output of the identified problem. If the distance from the center is equal to or greater than the threshold value for all three persons, the first evaluation unit 121 cannot identify which one has a problem, and outputs the fact to the first determination unit 131.

  In the processing cycle N, the second evaluation unit 122 compares Aro_N, which is the risk map 908 calculated in the processing cycle N, with PreAro_N-1_N, which is the output of the behavior prediction unit 113 calculated in the processing cycle N-1. . Specifically, for each object included in Aro_N or PreAro_N-1_N, a match determination using a predetermined threshold value is performed in the same manner as in the first evaluation unit 121. The second evaluation unit 122 may determine that the objects match as a whole when the number of objects determined to match exceeds the majority of the number of objects, or determine that the objects match as a whole only when determining that the objects match. You may. The second evaluation unit 122 outputs the result of the determination to the first determination unit 131. It can be said that the second evaluation unit 122 evaluates the consistency between Aro_N and PreAro_N-1_N.

  In the processing cycle N, the third evaluation unit 123 indicates Aro_N, which is the risk map 908 calculated in the processing cycle N, and the position of the own vehicle included in the output of the trajectory planning unit 114 calculated in the processing cycle N-1. Compare with PrePos_N-1_N. In other words, it is evaluated whether or not the vehicle has been controlled according to the planned trajectory 909 planned in the immediately preceding processing cycle N-1. This evaluation can be said to be an evaluation as to whether the position of the host vehicle and PrePos_N-1_N are consistent. Similarly to the first evaluation unit 121 and the second evaluation unit 122, the third evaluation unit 123 performs a match determination that allows an error using a threshold value, similarly to the first evaluation unit 121 and the second evaluation unit 122. When both match, the behavior prediction unit 113 evaluates that the own vehicle has been controlled according to the planned trajectory 909. The behavior prediction unit 113 outputs the evaluation result to the second determination unit 132.

  FIG. 9 is a diagram illustrating the determination content of the first determination unit 131, that is, the first abnormality determination result 921 output by the first determination unit 131 for each condition. The first determination unit 131 uses the output of the first evaluator 121 and the output of the second evaluator 122 to determine the first sensor 11, the second sensor 12, the integration unit 112, and the behavior prediction unit 113. Detects the presence or absence of an abnormality. FIG. 8 shows that the first abnormality determination result 921 is determined according to a combination of the output of the first evaluation unit 121 and the output of the second evaluation unit 122.

  As described above, the output of the first evaluation unit 121 is the result of the comparison between the synchronized first sensor data 901, the synchronized second sensor data 902, and the immediately preceding risk map 908A. The output of the first evaluation unit 121 indicates that only the synchronized first sensor data 901 does not match, only the synchronized second sensor data 902 does not match, only the immediately preceding risk map 908A does not match, all three match, and all three do not match. One of the five. The output of the second evaluation unit 122 is either the vehicle surrounding information 907 and the immediately preceding risk map 908A match, or the vehicle surrounding information 907 and the immediately preceding risk map 908A do not match.

  When the output of the first evaluation unit 121 does not match only the synchronized first sensor data 901, the first abnormality determination result 921 is determined by the first sensor 11 or the time synchronization processing unit 111 regardless of the output of the second evaluation unit 122. Is abnormal. When the output of the first evaluation unit 121 does not match only the synchronized second sensor data 902, the first abnormality determination result 921 is determined by the second sensor 12 or the time synchronization processing unit 111 regardless of the output of the second evaluation unit 122. Is abnormal. When the output of the first evaluation unit 121 does not match only the immediately preceding risk map 908A, the first abnormality determination result 921 indicates that the integration unit 112 has an abnormality regardless of the output of the second evaluation unit 122.

  When the outputs of the first evaluation unit 121 are all the same and the outputs of the second evaluation unit 122 are the same, the first abnormality determination result 921 indicates that there is no abnormality. When the outputs of the first evaluation unit 121 are all the same and the outputs of the second evaluation unit 122 are not the same, the first abnormality determination result 921 becomes abnormal in the comparison circuit or the majority circuit. When the output of the first evaluation unit 121 is not the same for all three persons, the first abnormality determination result 921 detects only the occurrence of an abnormality, that is, the abnormality cannot be specified, regardless of the output of the second evaluation unit 122.

  FIG. 10 is a diagram illustrating the determination content of the second determination unit 132, that is, the second abnormality determination result 922 output by the second determination unit 132 for each condition. The second determination unit 132 determines whether or not the vehicle surrounding information 907 is normal, in other words, whether or not the sensor fusion processing is abnormal, using the determination result of the first determination unit 131. If the second determination unit 132 determines that the vehicle surrounding information 907 is normal based on the determination result of the first determination unit 131, the second determination unit 132 changes the second abnormality determination result 922 according to the output of the third evaluation unit 123. That is, in this case, if the outputs of the third evaluation unit 123 match, the second abnormality determination result 922 indicates that the trajectory planning process has been completed normally, that is, the trajectory planning unit 114 is determined to be normal, and the output of the third evaluation unit 123 is If they do not match, the second abnormality determination result 922 indicates that there is an abnormality in the trajectory planning process, that is, the trajectory planning unit 114 has an abnormality. When the second determining unit 132 determines that the vehicle surrounding information 907 is not normal, that is, that there is an abnormality, based on the determination result of the first determining unit 131, regardless of the output of the third evaluating unit 123, the first abnormality determining result is obtained. 921.

  As described above, the second determination unit 132 can use the output of the first determination unit 131 and the output of the third evaluation unit 123 to detect the presence or absence of an abnormality in the trajectory planning unit 114 and the vehicle control based on the trajectory planning unit 114. .

  Note that, based on the two results of the first abnormality determination result 921 and the second abnormality determination result 922, when an abnormality is detected in any of the automatic driving ECU 2, the first sensor 11, and the second sensor 12, the mode transition unit According to 141, a process for shifting the process in the automatic driving ECU 2 to the degeneration mode is performed. That is, the mode transition unit 141 transitions the own vehicle to the degenerate mode when at least the evaluation result in the third evaluation unit 123 is negative. In the degenerate mode, for example, in the case of a system having a driver, an abnormality is notified to the driver, and control is transferred to the driver after continuing control for a certain period as necessary. In the degenerate mode, the speed of the host vehicle may be lower than in the non-degenerate mode, and a process may be performed that prioritizes stopping safely more than reaching the destination.

  The notification to the driver is executed, for example, by outputting an audio signal to a speaker (not shown) provided in the host vehicle via a communication interface (not shown) provided in the automatic driving ECU 2. The output of the audio signal to the speaker may be performed by the first determination unit 131 and the second determination unit 132 that have detected the abnormality, respectively, or a functional block for performing a notification, for example, a notification unit is newly provided to the notification unit. It may be executed.

  In the case of a system without a driver, a change in the control method including degeneration is performed to continue control on the system. As an example of the function degeneration, for example, a faulty part of the microcomputer 21 is separated and control is continued in a state where the control arithmetic processing is simplified, or the recognition result of the first sensor 11 and the second sensor 12 is partially or entirely invalidated. Is to be used. These functions may be reduced in accordance with a method defined in advance, or may be determined by calculation in the automatic driving ECU 2.

  The automatic driving ECU 2 may notify the driver only because the evaluation of the third evaluation unit 123 is negative. The notification may be performed by the third evaluation unit 123, or a functional block that performs the notification, for example, a notification unit may be newly provided and the notification unit may be executed.

  It should be noted that, in the operation of the microcomputer 21 described above, the vehicle surrounding information 907 calculated in the immediately preceding processing cycle is not verified. For example, verification of the vehicle periphery information 907, which is the output of the integration unit 112 calculated in the processing cycle T-1, is not performed in the processing cycle T. That is, using the first determination unit 131 and the second determination unit 132, in the processing cycle T, the verification of the action prediction unit 113 and the trajectory planning unit 114 in the processing cycle T-1 which is the immediately preceding processing cycle, and in the processing cycle T The verification of the integration unit 112 is performed, but the verification of the vehicle surrounding information 907 in the processing cycle T-1, that is, the verification of the integration unit 112 is not performed.

  Therefore, with only the above-described configuration, the first sensor 11, the second sensor 12, and the integration unit 112 cannot be verified only in the first processing cycle in which the automatic driving is started, and the subsequent action prediction unit 113 and trajectory planning unit The validity of 114 cannot be guaranteed. Therefore, it is necessary to separately verify the validity of the first sensor 11, the second sensor 12, and the integration unit 112 in the first processing cycle.

  As a method of this verification, for example, the following processing can be performed as an operation before the start of automatic operation. That is, at a certain time t0 and another time t1, the processing of the integrating unit 112 using the map information 13 detects landmarks around the own vehicle, and if necessary, considers the movement amount of the own vehicle to determine a match. Do. Thus, it is conceivable to verify the validity of the first sensor 11, the second sensor 12, and the integration unit 112 by configuring a temporal multiplex system.

  If the validity is verified, the automatic operation is started at time 1. If the validity is not verified, that is, if a mismatch is detected, the same operation is performed until the automatic operation start time defined by the system. It is necessary to repeatedly perform the detection and prohibit the start of the automatic driving process when the mismatch is continuously detected.

  With the configuration described above, it is possible to detect an abnormality that occurs in each of the sensor group 1 used in the automatic driving ECU 21 and the integration unit 112 and the behavior prediction unit 113 in the automatic driving ECU 21 without multiplexing them. This method is based on a comparison calculation process using maps or information obtained in the process of the main function processing of the automatic driving ECU, and simplifies the configuration of the automatic driving system and reduces the load of the verification calculation to achieve low heat generation automatic driving. Electronic control device is realized.

According to the above-described embodiment, the following operation and effect can be obtained.
(1) The automatic driving ECU 21, which is an electronic control unit, acquires information on the surroundings of the own vehicle as sensor information from the sensor group 1 for each processing cycle, integrates the acquired sensor information, and integrates the vehicle surrounding information for each processing cycle. 907, a trajectory planning unit 114 that calculates a planned trajectory 909 in which the vehicle travels in the future using the vehicle surrounding information 907, and a third evaluation that evaluates the reliability of the trajectory planning unit 114. Unit 123 (trajectory evaluation unit). The third evaluation unit 123 determines the position of the own vehicle in the second processing cycle that is a processing cycle after the first processing cycle in the planned trajectory 909 calculated by the trajectory planning unit 114 in the first processing cycle, that is, Using the planned trajectory 909A and the vehicle surrounding information 907 created by the integration unit 112 in the second processing cycle, the reliability of the trajectory planning unit 114 in the first processing cycle is evaluated. Therefore, the automatic driving ECU 21 can evaluate the reliability of the trajectory planning unit 114 with a low load without multiplexing computations and using additional sensors.

(2) The automatic driving ECU 21 estimates a future position of an object existing around the vehicle based on the vehicle surrounding information 907 in the first processing cycle, and a risk map 908 indicating the position of the object in the second processing cycle. , A second evaluator 122 (integrated evaluator) that evaluates the consistency between the last risk map 908A and the vehicle surrounding information 907 in the second processing cycle, and a second evaluator 908A and the second evaluator 908A. And a first evaluation unit 121 (prediction evaluation unit) that evaluates the consistency between the synchronized first sensor data 901 in the processing cycle and the synchronized second sensor data 902 in the second processing cycle. Therefore, the reliability of the evaluation target of the third evaluation unit 123 can be confirmed in advance.

(3) The sensor group 1 includes a first sensor 11 and a second sensor 12.
The first evaluation unit 121 evaluates the reliability by majority decision of the immediately preceding risk map 908A, the synchronized first sensor data 901 in the second processing cycle, and the synchronized second sensor data 902 in the second processing cycle. I do. Therefore, even if sensor information is input from only two sensors, it is possible to determine which one has a problem by majority decision using the time-series information.

(4) The automatic driving ECU 2 determines the reliability of the first sensor 11 and the second sensor 12 in the second processing cycle based on the evaluation result of the second evaluation unit 122 and the evaluation result of the first evaluation unit 121. A first determination unit 131 is provided. Therefore, the reliability of each sensor information output from the sensor group 1 can be evaluated.

(5) The automatic driving ECU 2 determines that the integration unit 112 or the trajectory planning unit 114 is abnormal when the evaluation result of the third evaluation unit 123 matches a predetermined pattern. Therefore, it is possible to prevent erroneous determination of an abnormality due to the influence of noise or the like, in other words, to prevent overdetection of the abnormality.

(6) The above-mentioned predetermined pattern means that the evaluation result of the third evaluation unit 123 is negative more than a predetermined number of times continuously. Therefore, the automatic driving ECU 21 can easily determine whether the determination of the third evaluation unit 123 is a predetermined pattern.

(7) Notify the occupant of the host vehicle that the evaluation result of the third evaluation unit 123 is negative. As described above, this notification may be performed by the third evaluation unit 123, or a notification unit that is a new functional block may be provided and the notification unit may be notified.

(Modification 1)
In the above-described embodiment, the trajectory planning unit 114 calculates the trajectory on which the host vehicle will travel in the future. However, the trajectory planning unit 114 may calculate the position of the own vehicle in each future processing cycle and output those positions to the lower ECU group 3. In this case, the command unit 25 may not be included in the automatic driving system S.

(Modification 2)
In the above-described embodiment, an example has been described in which all processes of the integration unit 112, the behavior prediction unit 113, and the trajectory planning unit 114 are verified without a redundant configuration, but the processing load and the sensor configuration required for the automatic driving ECU 2 are described. In some cases, multiplexing may be allowed for only some sensors or processes due to other system requirements. In this case, the allowed configuration may be made redundant, and some of the above-described configurations may not include the evaluation unit.

  For example, when multiplexing of the first sensor 11, the second sensor 12, and the integration unit 112 is permitted, the first evaluation unit 121 is unnecessary. In this case, the behavior estimation unit 113 can be verified by the second evaluation unit 122 with the vehicle surrounding information 907 output by the integration unit 112 in the processing cycle N as the correct answer. Similarly, the first evaluation unit 121 is not required even when multiplexing of the behavior prediction unit 113 is permitted. In this case, the risk map 908 in the processing cycle N + 1 output by the behavior prediction unit 113 in the processing cycle N is regarded as the correct answer, and the second evaluation unit 122 verifies the vehicle surrounding information 907 output from the integration unit 112. be able to. When multiplexing of the trajectory planning unit 114 is permitted, the third evaluation unit 123 is not required.

(Modification 3)
In the above-described embodiment, an example has been described in which the first evaluation unit 121, the second evaluation unit 122, and the third evaluation unit 123 are independently mounted. However, the first evaluation unit 121, the second evaluation unit 122, The function of the third evaluation unit 123 may be integrated. That is, it is sufficient that the determination shown in FIGS. 9 and 10 can be made, and the function allocation and configuration are not particularly limited.

(Modification 4)
In the above-described embodiment, the abnormal part is determined by comparing the calculation results in the two processing cycles of the processing cycle N and the processing cycle N + 1 that is the next cycle of the processing cycle N. This is merely an example, and the abnormality may be determined by comparing the position information, the predicted position information, and the predicted trajectory information over a plurality of times, for example, at times T, T + 1, and T + 2. Further, the processing cycles to be compared do not necessarily have to be continuous. For example, an abnormal part may be determined by comparing the calculation results in two processing cycles of the processing cycle N and the processing cycle N + 2. Further, an abnormality may be determined based on abnormality detection over a plurality of times, for example, five or more consecutive detections of the same object or location, or a temporary change caused by a sudden change in sensor characteristics, noise, or surrounding environmental conditions. May be excluded.

(Modification 5)
Each of the first evaluation unit 121, the second evaluation unit 122, the third evaluation unit 123, the first determination unit 131, and the second determination unit 132 keeps a record when a negative evaluation or determination is performed. Is also good. Further, each of the first evaluation unit 121, the second evaluation unit 122, the third evaluation unit 123, the first determination unit 131, and the second determination unit 132 determines that the evaluation result is negative when the evaluation result is negative. The evaluation result may be recorded for a predetermined time before and after the determined time as a starting point, for example, one minute before and after the time when the negative evaluation was performed. When recording the evaluation result, the time and the input data value may be recorded together.

  A negative rating is anything other than a positive rating. For example, the first evaluation unit 121 outputs a positive evaluation only when the output of all three matches is a positive evaluation, and otherwise outputs a negative evaluation. In the second evaluation unit 122, a match is a positive evaluation, and a mismatch is a negative evaluation. The third evaluation unit 123 is a positive evaluation with a match and a negative evaluation with a mismatch. The first determination unit 131 has a positive evaluation if there is no abnormality, and a negative evaluation otherwise. The second determination unit 132 has a positive evaluation that the trajectory planning process has been normally completed, and a negative evaluation otherwise.

(8) Each of the first evaluation unit 121, the second evaluation unit 122, the third evaluation unit 123, the first determination unit 131, and the second determination unit 132 records the evaluation result that the evaluation result is negative. . Therefore, the record can be confirmed after the fact.

(9) When each of the first evaluation unit 121, the second evaluation unit 122, the third evaluation unit 123, the first determination unit 131, and the second determination unit 132 determines that the evaluation result is negative, the evaluation result is negative. The evaluation result at a predetermined time before and after the time determined to be the target is recorded. Therefore, a large amount of information is recorded, which can be used for subsequent analysis.

  The above-described embodiments and modifications may be combined with each other. Although various embodiments and modified examples have been described above, the present invention is not limited to these contents. The above-described embodiments and modified examples are merely examples, and the present invention is not limited to these contents as long as the features of the present invention are not impaired. Other embodiments that can be considered within the scope of the technical concept of the present invention are also included in the scope of the present invention.

1: Sensor group 2: Automatic driving ECU
11 first sensor 12 second sensor 14 own vehicle position sensor 21 microcomputer 41 time synchronization processing 45 vehicle peripheral information 111 time synchronization processing unit 112 integration unit 113 behavior prediction unit 114 track planning unit 115 ... Behavior prediction storage unit 116 trajectory plan storage unit 121 first evaluation unit 122 second evaluation unit 123 third evaluation unit 131 first determination unit 132 second determination unit 141 mode transition unit 901 first Sensor data 902 second sensor data 907 vehicle surrounding information 908 risk map 908A last risk map 909 planned trajectory 909A last planned trajectory

Claims (12)

An integration unit that acquires information around the vehicle as sensor information from a plurality of sensors for each processing cycle, and integrates the acquired sensor information to create vehicle surrounding information for each processing cycle;
A trajectory planning unit that calculates a planned trajectory in which the vehicle will travel in the future using the vehicle surrounding information for each processing cycle,
A trajectory evaluation unit for evaluating the reliability of the trajectory planning unit,
The trajectory evaluation unit, in the planned trajectory calculated by the trajectory planning unit in a first processing cycle, a position of the vehicle in a second processing cycle that is a processing cycle subsequent to the first processing cycle; An electronic control device for evaluating the reliability of the trajectory planning unit in the first processing cycle using the vehicle surrounding information created by the integration unit in the second processing cycle. The electronic control device according to claim 1,
Behavior prediction for estimating a future position of an object existing around the vehicle based on the vehicle surrounding information in the first processing cycle and creating a risk map indicating the position of the object in the second processing cycle Department and
An integrated evaluation unit that evaluates consistency between the risk map and the vehicle surrounding information in the second processing cycle,
An electronic control device further comprising: a prediction evaluation unit that evaluates consistency between the risk map and the sensor information in the second processing cycle. The electronic control device according to claim 2,
The plurality of sensors include a first sensor and a second sensor,
The prediction evaluation unit is configured to evaluate reliability based on a majority decision of the risk map, the output of the first sensor in the second processing cycle, and the output of the second sensor in the second processing cycle. Control device. The electronic control device according to claim 3,
Electronic control further comprising a first determination unit that determines reliability of the first sensor and the second sensor in the second processing cycle based on an evaluation result of the integrated evaluation unit and an evaluation result of the prediction evaluation unit. apparatus. The electronic control device according to claim 1,
An electronic control device that determines that the integration unit or the trajectory planning unit is abnormal when the evaluation result of the trajectory evaluation unit matches a predetermined pattern. The electronic control device according to claim 5,
The electronic control device, wherein the predetermined pattern is that the evaluation result of the trajectory evaluation unit is negative more than a predetermined number of times continuously. The electronic control device according to claim 1,
An electronic control device further comprising a notifying unit that notifies a passenger of the vehicle that the evaluation result of the track evaluation unit is negative. The electronic control device according to claim 1,
The electronic control unit, wherein the trajectory evaluation unit records the evaluation result when the evaluation result is negative. The electronic control device according to claim 8,
An electronic control unit that, when the evaluation result is negative, records the evaluation results at a predetermined time before and after the time when the evaluation result is determined to be negative, as a starting point. The electronic control device according to claim 1,
The electronic control device further includes a mode transition unit that transitions the vehicle to the degeneration mode when the evaluation result by the track evaluation unit is negative. The electronic control device according to claim 1,
An electronic control unit that ensures the reliability of the plurality of sensors and the integration unit by redundancy. The electronic control device according to claim 1,
An electronic control unit, wherein the plurality of sensors include at least one of a radar and a camera.

Images